Re: Kosko considered fuzzy (Re: [HELP] Fuzzy Neural N

Bogdan Kosanovic (bogdan@neuronet.pitt.edu)
28 Mar 1995 18:54:01 GMT


In article <1995Mar28.123021@GenRad.co.uk>, browne@GenRad.co.uk () writes:


> Seems I've opened a can of worms here... Everyone agrees that Kosko is the
> main man in the field, but I'm sure we have all read technical manuals
> that were written by experts before. And they are invariably written assuming
> that one understands the subject already. I'm not defending Kosko's book
> as I haven't read it yet, but it seems that the griping should be replaced
> by personal interpretation, so that simpletons, like me, can get something
> constructive from the conversation. ;-}

I did not have time to follow this discussion, but would like to add my
own experience with the book ("Neural Networks and Fuzzy Systems"):

- it helped me better understand how neural networks internally work,
- it helped me understand what are the fuzzy sets and how they can
help me in my research (signal processing, systems theory)

I am not interested in control theory, so I cannot comment on that part
of a book.

Overall, I'm happy to have that book on my shelf. To fully appreciate its
worth, one needs to have a bit more knowledge of math (basic topology,
functional analysis), although it is not necessary.

For full coverage of fuzzy sets and systems I also recommend a very good
book of Dubois and Prade: "Fuzzy Sets and Systems: Theory and Applications".
This book contains an excellent review of a variety of "fuzzy" material.
Some chapters require knowledge of advanced math.

At last, but not least ;-), a book of Klir and Foger is very good for those
interested in modeling uncertainty: "Fuzzy Sets, Uncertainty, and Information"

The above three books together may provide very good source of information
for those of you who are interested in staring research related to
fuzzy sets, fuzzy logic or fuzzy systems. (I guess there are also books on
fuzzy control, but I did not pursue that line of research...)

Regards,
Bogdan R. Kosanovic

-- 
Laboratory for Computational Neuroscience    e-mail:  bogdan@neuronet.pitt.edu
University of Pittsburgh                      phone:  (412) 624-6773
http://www.neuronet.pitt.edu/~bogdan/

------------------------------